New World Newsfeed: Second Life's Growth Plateau Ends

GigaOM: Exclusive: Second Life Starts To Grow Again

This morning I filed a GigaOM story based on an internal data chart from M. Linden, showing real substantial growth to Second Life's user base since last August, now registering 731K monthly unique repeat log-ins (but not counting new accounts created during each month.) See the full chart here.

Currently, the Lindens report monthly active users (defined as Residents who log in for at least an hour a month) as 650K last March. That's roughly a 100K increase from the average of the last couple years, when monthly active users were mired somewhere around 550K. In other words, the growth plateau that has plagued Second Life since mid 2007 finally seems to be at an end.

M. Linden tells me he considers monthly unique repeat log-ins (or MURL) as an equally meaningful metric to monthly actives, if not more so. Why?

Because the monthly active user rubric includes the many new users who spend more than an hour in the orientation process, but still eventually quit out and never return. The MURL metric doesn't count them at all. "These are log-ins not people who spent one hour in-world," as M. told me by email, "very different threshold (and probably more typical of what you see from other social media sites.)" Now, he continued, "The figure is 731K, which is up from 582K a year ago."

Yes, he acknowledged, the 731K figure does include bots; however, he tells me that aggregate bot activity has remained consistent through the last couple years. (He affirmed Zee Linden's estimate that 10% of user activity are actually bots.) In other words, notwithstanding bots, he believes most of the growth still comprises real users. Moreover, the user base is spending Linden Dollars. "In March, 45% of Residents spent L$. That figure has been stable for some time." As he put it, "I doubt bots do that."

Looking at the MURL growth rate now, the number may surpass a million by the end of 2009. Whatever the case, it seems undeniable that Second Life's plateau period of its middle years has officially ended.

Comments

[Adric Antfarm]

Do we know of this number how many are subscribers?

[Adric Antfarm]

Also (and no, I am not writing a book) who has a good source for places people honestly go to and like.

I started with b.places HUD that sent me to places that looked popular in 1973, but seemed kinda abandoned now (it this still maintained?)

Then I found a site ranking "top sims" which in truth only ranks people who click on a vote box which is more a click game really. When I see a donation box (like at a range the other day) and I'm getting to do something like shoot for free, I always drop something in, but I know I am in the minority whereas a free box marked VOTE simply has to be shinny and need people with ADD to visit the sim.

And yes - I know this very site ranks yearly and perhaps more often sims. This is however a rank on the most creative/cool/appealing sites whereas I want to know where the people are going and like.

If you have a lot of time on your hands you can zoom out on the map and find the clusters which always leads me to the same findings. "Free Money Place / Island / Tree" are the top sims. This is a shock to me as I was sure it was more naughty than that, but as I said - I am looking for more resources.

I would like an in game search tool (and SL "Recommend" tab that is useful instead of lame) that is less open to keyword gaming. A search for say "slavery" isn't listing sims on the injustice of past it's -- well, you know.

What really p's me off is searching for the word "rape" (even with mature off) shows a rape crisis item as the 2nd. The 1st? Hamlet would give the Two World shove (and properly) for telling you what the 1st reads. It's offesive beyond all means (and I am a pretty open minded guy). Imgine if I was someone online looking to talk to someone about something that happened to them? How does seeing that do anything but fark with their head all the more?

[Hamlet Au]

Hold that last thought for an open forum, Adric. Far as subscribers, that's definitely a good question for me to ask in a follow-up. I imagine the Lindens will downplay it, since it's been dropping and they're talking about making it a more attractive offer.

[Adric Antfarm]

I knew it. I heard we would get bumped to $302.50 a week to make us less resentful of the those grandfathered in at $500.

Finally. Justice.

[anna gulaev]

So, they're going to stick to the 10% bots number? They aren't that stupid. They must think we are. I'm not flattered.

[Luce Imaginary]

@anna gulaev, could you present your data and the methodology you use to get your estimate of bot population? It would make for an excellent paper.

[benjamin s]

Is the economic downturn spurring some of this growth? i.e., more people with (a) increasing free time due to underemployment, and (b) stronger desire to imagine a different reality?

Has anyone cross-indexed these numbers with unemployment rates? I suspect a correlation... but wonder at its strength.

[M Linden]

@Adric -- we hear you on Search/Showcase, etc. The whole find and discover toolset is getting an overhaul. Also, we do know how many are subscribers. I got whomped once before when I said premiums are not material to our business. What I meant was the revenue from the premium subscription itself wasn't material because we have other larger revenue streams. What I should have said was that Landowners are vital to our business and when we lose Landowners it hurts...not because of the loss of the premium membership fee, but because we are losing a valuable economic participant. That being said, we have a lot of new land buyers every month which is a really encouraging trend. And, even with some softness in Land maintenance and declining premiums our other revenue streams more than compensated for that softness.
@Hamlet -- Hi James :) Surprised I'm here? We won't downplay the dialog on Premiums, if it's of interest. The reason we react strongly when people say "premium memberships are declining therefore Second Life is dying" is because that's just plain wrong. And we will get premium membership back on a growth trajectory. Not tomorrow but certainly within the next 6-12 months. T Linden is doing a Q1 results post later today on the Second Life blog. It will address some of the questions folks have.
@Benjamin -- undoubtedly the economic downturn has contributed to the growth in user hours. As for growth in new users, partially. One of the key drivers of user growth is that we have been making it easier for people to join Second Life (simpler reg form) and that the platform is much, much more stable than a year ago. Yes, we still have a lot of work to do, but we are making progress.
@Luce -- I don't think any of our Residents are stupid. Far from it! There are definitely bots in SL. I've met some myself :) Some are quite useful. Some are not. Whether or not you agree with our estimates (10-15%), there is no doubt we are adding real new users at an increasing rate and that's what's most important. We all agree bots exist and some are not good. Jack Linden has been blogging on this.

[Sioban McMahon]

"And we will get premium membership back on a growth trajectory."

Take a look at what premium membership buys your customers/residents. Make the pot a bit sweeter, for a start. I think premium membership indicates a stronger commitment to participation in SL. You want to move more folks over into that category, certainly.

Making free memberships easier to set up has it's ups and downs. Of course, helps make a person's first experience with SL less intimidating and reduces the likelihood that they'll give up on the whole thing. Keep in mind, also, though, that easier sign up will help the folks creating alts for mischief purposes.

[Marianne McCann]

What Sioban said. Consider added features for those who feel it appealing to pay into Second Life with a premium membership and land ownership. What features bonuses would premiums want? Additional location accesses? I know I'd love to see additional group spaces available for premiums, and would think this a "value added" item. Others might want some additional land controls, or maybe even "early previews" of Linden lands (Ursula City, etc.). I'm sure there are others.

[CT Xue]

I'm actually interested in purchasing roughly a 9-region lot and committing to paying the monthly tiers for quite a while, but as a business owner myself I don't like the half-hearted attempt to fix the grid performance this year, so I wait... Add-ons and value-added services would be a great option! If I want better performance, more scripts, more prims, less lag, etc...Tell Me What It Costs, and I'll write the check. For now, I am not going to expend any monies at all into something I have no control over. If Linden Labs wants more users, get someone like me excited about it and we'll rock the joint! Server asset use was protected by reducing allocation to sims, thus not needing an infrastructure investment, and prices were raised to compensate for the loss of clients we knew would happen. Net-net is the same, but I still haven't seen what I want to buy yet.

[M Linden]

@Sioban, Marianne, CX -- you all have very valid points about adding value to subscriptions and we are ON IT BIG TIME :) There is so much good stuff on the way...not as quickly as we or you would like but it is definitely on the way. We've done some research to see what people value and we'll do more but your suggestions are terrific. Thank you!

[Sioban McMahon]

M,

Thank you for coming out to the wilds of the web to chat with us. It's remarkable that someone at your level is accessible. We've even had you visit our Pub, Nancy Blake's, in the West of Ireland. I have to admit that most folks were dumb struck when you tp'ed in. We grateful for our donation to our charity, too.

[M Linden]

Thanks Sioban. I like getting out and about in SL!

[Koinup Burt]

@Adric We at Koinup deployed a service called: "Second Life Place":
http://www.koinup.com/places

Launched some months ago, it is quickly becoming one of the most popular resources for discovering new places in SL.

The ranking is based on the number of people that shared/photographed that particolar slurl on Koinup. So the most popular on the "Second Life Places" section are the most popular amont the community.

I invite you to have a look, try it and leave us a feedback. Btw, New World Notes is media partner for Koinup and weekly it's featuring on this blog the most coolest places

[Henri Beauchamp]

How M. Linden is to explain that the "Logged In Last 60 days" figure (to be seen on the login screen of the viewer) has grown quickly till November 2007 to peak at a little over 1 500 000 users and then stagnated below this level ever since ?

The adoption of a new indicator that *appears* to show more favourable figures is in fact just a way to try and hide the *truth*. The truth is that the *actual* number of *active* users (and this shall exclude bots and campers which are certainly way more than the quoted 10% figure) has been declining in 2008 and stagnating since.

M. Linden is just spreading corporate lies to try and hide the failure of Linden Lab, failure which can very easily be explained with all the catastrophic decisions the lab made in 2007 (adoption of the new, buggy web-based search engine, abandon of the legacy renderer for the viewer, letting 30% of the users behind), 2008 (Open Space sims amputation), and 2009 (absorption of OnRez and XStreet and destruction of the former, leading to a monopoly, and soon segregation of so-called "adult" activity to please the bigots).

Congratulation, Linden Lab ! You turned a dream world into a nightmare.

[Hamlet Au]

Henri, those are serious charge to make, and I'd prefer you keep things civil, please. What's your evidence that active users are below what is reported?

[Henri Beauchamp]

@Hamlet

The evidence is right under everyone's nose when logging on on SL, namely the "Logged In Last 60 days" figure in the upper right corner of the viewer login screen.

The peak number was reached on 2007-11-09 with 1512924 users (see: http://s3.amazonaws.com/static-secondlife-com/reports/marketplace_stats/2007-11-09/logged_in_users.xml ), and has been less since this date.

Former stats used to include this figure in the graphs. Now, they replaced it with another, better looking (but mostly irrelevant) indicator, to try and give false proofs that the growth is back.

I'm also quite upset that Linden Lab pretended that they ran a survey among all the merchants to know how was their business going, while all the merchants I could speak with were not at all asked about this by LL (and they did not ask me either). I suspect that LL selected the (few) merchants which business was growing or at worst stagnating, purposely ignoring all the others in an attempt to publish more corporate lies.

The truth is that the revenues of the large majority of the merchants decreased in large proportion over the last two years, while the costs for running their business increased (and is going to double because of soon to come the "adult" stuff segregation: merchants who, like me, offer both "adult" and "mundane" products will have to split their shops in two, doubling their costs).

The reasons are many (the buggy search engine is one, and dates back from the end of 2007 when the viewers lost the "All (old)" search tab), but an obvious reason is that the user base has been stagnating for the last two years, and this mainly because of Linden Lab's very poor strategy and catastrophic decisions.

Seeing them now congratulating themselves is really getting on my (and many other residents') nerves. Do they really take us for stupid and blind persons ?

I will only grant one thing to LL: the grid and viewer stability are much better... But all the rest is getting worst with time, I'm afraid.

[Hamlet Au]

Henri, you make a good point, and I'll bring it up with M and T. At the same time, I think it's way too early to conclude the new numbers are deceptive. Recall that the MURL metric reflects unique *returning* users, while the "Last Logged-In" metric includes new sign-ups, most of whom churn out. It could be the apparent flatness of log-ins is really just reflecting that "one try then bye bye" process. Also, assuming flat growth, I'm not sure how to explain the consistently rising concurrency. Yes, bots represent some of that total, but I'd want to see compelling evidence they're growing or represent a larger percent than %10 total activity.

[Henri Beauchamp]

@Hamlett

Concurrency increases (albeit much slower than it used to do back in 2006, when it increased exponentially), but it's only the result of the *same* people (and bots, and campers, which IMHO are closer to 30% than 10-15% as quoted by M.) spending more time online on SL.

However, the "Logged In Last 60 days" is a much better and way more relevant figure, since it smooths out the various spikes (such after an interview about SL in the media) of new registrations, which are followed a few weeks later by a larger amount of "leavers".

M. is true with the argument that SL is easier to get in the first time, but I'm afraid the rate of the newbies who register then leave SL only a few days later has only been growing to compensate the new influx of users resulting from the easier registration process.

The *actual* user base, is very well reflected by the "Logged In Last 60 days" figure, and I always could see a close correlation between this figure and my revenues as a merchant.

Another figure that is relevant (and on the down slope) is the number of premium accounts: those accounts correspond to residents who actually invest money in SL.

No need to be an expert in statistics to see that LL's conclusions do not match the cruel truth.

As for declining businesses, there is also another effect: more and more residents are opening businesses (thus the growing numbers of exchanged L$), but since the user base is the same, and that most of the residents pertaining to it already got almost all they need for their aging Avatar(s), the revenue per business will at best stagnate, and in most case go down. Only an increasing number of active users could get us back to the levels we experienced back in 2006...

[Hamlet Au]

"the 'Logged In Last 60 days' is a much better and way more relevant figure, since it smooths out the various spikes (such after an interview about SL in the media) of new registrations"

I'm not sure I follow you. Can you explain how a number that includes two months worth of noob churn is more meaningful than a number than only counts users that have been in-world at least twice?

"Concurrency increases... but it's only the result of the *same* people"

Henri, what's your evidence for this?

[Henri Beauchamp]

An averaged figure is always best at giving true tendencies: spikes are smoothed out. It suffice that some media (TV, radio) speaks about SL, and you will get a huge spike in newbies, but a very large majority of them will most likely not return (SL is still very "geeky" and hard for the non-techies to deal with).

Plus, the MURL is a joke: what happens if a newbie crashes and relogs (or when many rolling retarts happen in the week) ?... Will it be counted as a "unique repeat login", even if they never return during the month ?...

As for the concurrency, it's pretty obvious. See the charts (login time increasing per resident = more concurrent connections per hour).

[Hamlet Au]

"what happens if a newbie crashes and relogs (or when many rolling retarts happen in the week) ?... Will it be counted as a 'unique repeat login'"

Presumably that would be counted as a repeat log-in. However, as you just said, stability has been improving in recent months. So repeat log-ins due to crashes are probably going down, not up. Therefore I still don't see why MURL is a less meaningful figure.

"See the charts (login time increasing per resident = more concurrent connections per hour)"

What chart are you referring to?

[Henri Beauchamp]

MURL: crashes are not that rare (especially with the official viewer), even if their rate is much lower than before... And what if a newbie logs off and back on (for whatever reason, including RL interruption, Windoze going unstable, etc) then never return ?...
Instead, the "Logged in last 60 days figure" gives you the "sum of new users + stable user base", but with the "spikes" smoothed out and averaged over two months.
In any case, I'd really like to hear (or rather read) the explanations of M. Linden on why the "Logged in last 60 days figure" is stagnating when he pretends the number of active users increased...

The charts are here:
https://blogs.secondlife.com/community/features/blog/2009/04/16/the-second-life-economy--first-quarter-2009-in-detail

Have a close look at the "User hours per quarter" and "Peak concurrency users in the quarter" curves, and see how they are almost perfectly parallel... It is the proof that the increase in the peak concurrency does *not* reflect a larger amount of active users, but just the consequence of having users spending more time online...
And guess what: I do believe (although I have no proof), that the increased amount of online time per user is actually the result of having more and more of bots/campers online. Why ? Because in the last months, the offering for bots software and lightweight (non graphical) clients (such as OMV-light) has enlarged, making it possible for people to stay camping with several avatars logged in at once, either to make money for themselves out of the camping pads, or to artificially increase the "traffic" figure of their parcels/sims (1 avatar logged 24H a day = 1440 added to the traffic)...

[Hamlet Au]

"Instead, the 'Logged in last 60 days figure' gives you the 'sum of new users + stable user base', but with the 'spikes' smoothed out and averaged over two months."

I'm not sure that's always the case, Henri, new user spikes happen often and erratically, with several that sometimes occur within a sixty day period, followed by relatively few spikes during the next sixty day period. Especially within the time frame you're looking at, of November 2008. That month, as you may recall, several UK publications reported on a virtual adultery story that caused an enormous spike; the story was subsequently picked up by the US press the next week, and that caused *another* spike. I talked with a Linden recently who told me they'll often see a new user spike literally going across the world as the same media report gets syndicated and slowly goes global over a few weeks or even months. So I'm not confident that the 60 day log-in metric averages out those spikes of new users who churn out.

"Have a close look at the 'User hours per quarter' and 'Peak concurrency users in the quarter' curves, and see how they are almost perfectly parallel... It is the proof that the increase in the peak concurrency does *not* reflect a larger amount of active users,"

That simply doesn't follow. It may be the case as you say that rising concurrency is mostly established users logging in more; it may equally be the case the data suggests, as the Lindens claim, that it represents a growth in user numbers *and* user hours. I would need to see more evidence to be convinced of your interpretation.

For months I've been skeptical that Second Life was growing, and this internal data chart that I posted on GigaOM came after a long conversation I've had with M. going on several weeks, constantly asking for proof that SL's growth plateau had ended. Until I see really substantial evidence otherwise, I have to conclude that's indeed the case.

Speaking of which, I very much welcome a well-sourced third party study of SL user behavior with good methodology (I'm surprised Metanomics or another such outfit hasn't done this yet), but until then, we're really just left with the data the Lindens offer, and our own common sense.

[Henri Beauchamp]

You are just giving more arguments in favour of a stable user base: I was not aware of the publicity made around SL in UK in US (2008 has been pretty void of SL-related news, here, in France, unlike 2006 and 2007).

Since, according to the info you are giving, the spikes took place starting from November 2008, then it indeed explains why the "Logged in last 60 days" figure experienced a 15% or so increase between October 2008 (with figures oscillating between 1048034 and 1294493 users) and January 2009 (between 1274403 and 1461324 users).

As I already said, the "Logged in last 60 days" is made of the addition of the stable user base and the newbies. Since there has been more newbies at the end of 2008, the stable user base staid pretty constant ("Logged in last 60 days" was almost constant and around 1200000 from January to October 2008).

Now, and since the "Logged in last 60 days" figure is still around 1450000 right now, it looks like either some of the newbies from those late 2008 spikes became regulars, *or* an increased amount of newbies are coming to SL but the user base is still stable... Only time will tell.

In any case, it is MUCH too early for Linden Lab to brag about the (hypothetical and still unproved) return of the growth of its user base. Let me recall you that the all times record for the Logged in last 60 days" figure was in November 2007 with 1512924 users, and we are still *BELOW* this number right now...

[M Linden]

@ Hamlet -- thank you for suggesting common sense prevail in this dialog :)

@ Henri et al -- there is a lot of back and forth in the commentary that is confusing the issue.

First off, there is a big difference between 30/60 day log-ins and Monthly Unique Repeat Log-ins (MURLI). What is the difference? The former counts ALL log-ins (not just Repeat users). MURLI counts precisely what is in the name. Unique Repeat Log-ins. It doesn't count one-timers, it counts REPEAT users.

30/60 day log-ins are pretty much always larger than Monthly Unique Repeat Log-ins (MURLI).

Why is 30/60 day log-ins a less accurate measure of growth? It includes "one-timers" -- new users who sign up, log in and leave, never to return.
From a press-coverage perspective, Second Life was the Twitter of 2007. During that hype phase, a lot of people signed up, logged in and left. 2007 had a lot of one-timers which is why log-ins were high.

Because 30/60 day log-ins are subject to "bounce" from one-timers we don't look at 30/60 day log-ins as a real indicator of whether or not we are growing our active user base.

Instead we look at Monthly Active Users and Monthly Unique Repeat Log-ins. Both those measures show we are growing. And, when you look at the MURLI diagram Hamlet posted, you can see that we have been growing continuously since we launched. Period.

Now let me address Henri's point about viewer crashes causing repeat log-ins. All I can say is, that's a stretch. Our crash rate has been declining so you can't attribute our growth in repeat log-ins to an increasing crash rate. That just doesn't make sense.

Next, let me talk about peak concurrency. It too has been growing. Hooray! But, you have to be careful when comparing that to user growth because peak and trough have varied over time depending on which time zones we are growing in. Even in my 11 months here, the shape of the concurrency curve (not just the height) has changed considerably.

Finally, re: bots. Sure, there are bots. Our Residents are creative in how they're used and not all are used for good. The last analysis we did showed them at around 10% which is the figure Z quoted and Hamlet asked me to confirm. When we have a new analysis, we'll share it. What I can say is that all the evidence shows that our REAL user base is growing and that it isn't just bot-mania.

Linden Lab has hundreds of metrics on our internal dashboard. I watch them closely every day. We have a large data warehousing team that responds in an instant when there is a data anomaly so we are very knowledgeable about our data. I can look at any number of measures and they say the same thing -- the Second Life user base is growing and Second Life economy is growing. Period.

Thank you for your intensity and passion. See you in Second Life.

Cheers,

M

[Henri Beauchamp]

@M. Linden

Thank you for your reply.

However it sounds to me like you are using the "méthode Coué" (repeating the same thing over and over to try and convince people).

Fact is that you (well, not you, but someone at LL, even before you became the CEO) deprived us from another useful figure which would have helped a lot to make a clear distinction between the one-time residents (the newbies who come one day (or one week) and never return), and the actual "residents" (the stable user base): this figure (and associated graph) is the total number or Second Life "users" (actually, registered avatars). It used to be shown on the login screen, together with the "Logged in last 60 days" figure... It disappeared back in 2008 (or was it 2007 ?).

As for MURLs, I am sorry, but your assertion following which it is a reliable indicator is completely wrong: even a "not to return" newbie will have to log off and back on at some point. Unless they are really completely baffled by SL, they will most likely have a few tries (perhaps even over a full week) before they decide SL is not for them and not return. MURLs will count such avatars as part of the stable user base while they are not.

There is also the issue with rolling restarts, asset server issues, failing TPs getting you logged off, etc... I counted no less than 15 such issues for March 2009 (and only counting issues affecting the whole grid) in the Second Life status blog (and I'm afraid they are not *all* listed here: I can't count the times when the asset servers got short glitches that force you to log off and back on).
During such issues, pretty much all the online users have to log off and back on at some point. With an average concurrency in March somewhere around 50000, that's a lot of MURLs added to your stats (since pretty much everyone online will have to relog at one point or another), but not to the (still more reliable, IMHO) "Logged in last 60 days" figure. I could also add the random sim crashes (or sim server restarts by the estate manager), the avatars stuck in an anim or missing/corrupted textures (forcing them to relog), etc, etc...

I'm sorry, but the way your MURL figure is defined does not make it a reliable way of measuring the number of residents in the stable user base.

There is also the issue of the alts... Back in 2006, residents had to pay a $10 fee for every new alt they registered... Back at that time, I got only one main avatar and one alt. After alts creation became free (sometime in 2007: I honestly can't remember the exact date for this change), I made two more alts, meaning I appear as 4 distinct users in your MURL stats... and I know people on the grid with more than 10 alts, so I'd say I'm just in the average with my 3 alts (+1 main avatar).
Don't you think it is therefore hard to compare the figure of 2006 with the one of 2008 because of this very change ?

Now, you definitely *could* give non disputable figures: the viewer sends to LL the MAC address of the network interface it is connected through. This MAC address (unlike the IP, which can change over sessions, depending on how your ISP attributes the IPs) is a pretty reliable characteristic (identifying the computer of the user itself) that could narrow the error margin dramatically when counting the number of *unique users*, even getting rid of the fake bloat in the stats which is due to the alts.

You see, I would be ready to believe you, but only after you give *relevant* and *coherent* figures, and I'm afraid it's not the case for now. As a further proof of a stagnating user base, *I* (and many, many merchants I spoke with) have a more reliable figure than you do: their revenue at best stagnated, or decreased over the 2006-2008 period, and 2009 is not showing a reversal in the tendency (though I will grant you that the Real Life economic crisis is certainly a factor).

So, please, M., before bragging about how wonderful is the growth of SL, be more careful and bring non disputable proofs (which I'm sure you could give).

As a last note: please, pretty please, I beg on my knees... change your mind about the adult segregation. Implementing it would be a HUGE and TERRIBLE mistake that will only bring our businesses further down into the abysses and that will ruin the experience of 90% or your residents, in turn impairing the growth of SL as a whole.
If you really want to segregate some people, open a PG-only continent, move all the corporate, education and bigots over there, create a user setting for restricting access to this PG land onlmy (i.e. if the *user* chooses so, they set themselves as PG only and get protected from "adult"/"not work safe" contents), and let us, normal people, enjoy SL as we always did !
Many thanks in advance !